Analyzing dynamic risk of stampede in stadium: A quantitative method considering the various status of risk factors in whole process

Abstract

With the increase of large-scale mass activities, stampedes of deaths and injuries easily occur. The bidirectional qualities between the state's shifting combinations of different risk factors and accidents must be more thoroughly analyzed. The triangular fuzzy number and the Bayesian network are applied in the study to construct a risk analysis model for stadium stampede accidents. 46 accident samples and 24 risk factors closely related to activities were collected for the study. The results show that 7 factors, including audience mood, vulnerable groups, retrograde crowd, abnormal person, flammable and explosive material, walkway in stadium, and special time period, are more likely to cause stampede incidents. When the factors related to the live audience are added together in the most dangerous state, the probability value of the activity safe state decreases from 86% to 31%. In particular, the mood state of the audience, ranging from the more excited to crazy, is most sensitive to stampede. In addition, four major causal chains are obtained. The two risk factors, crowd flow limitation measures and safe entrance and exit, contributed most significantly to the stampede when the personnel volume ratio began to exceed 80% or when the security entrances and exits were partially closed. The model constructed in this paper can dynamically demonstrate the rules between the transition combinations of different states of risk factors and stampede. It will provide some theoretical basis for the process safety management of large-scale mass activities in stadiums.